Image prompt for Bing/DallE: a doctor, a nurse, a pharmacist, an IT guy work together to improve the lives of doctors who are using computers with patients, optimistic, impressionistic, colorful
Our EMO committee (electronic med-alert optimization) is looking to slash medication alerts dramatically. We asked our pharmacists to look at the current alert settings on “lactation warnings” (which of these medications should not be taken during breast-feeding) and weigh in on which could be turned off. For example, we had a number of obstetrics/gynecology professors tell us that “amoxicillin is safe during breastfeeding, please turn of this alert, it is not accurate.”
First of all, EMO: coolest committee name ever.
Secondly: how hard can this be?
When we gave this and similar requests to our clinical pharmacists, they were wary of removing ANY alerts.
“Yes amoxicillin is low risk, but it is not ZERO risk. We should still show this alert. Besides, I don’t want to be responsible if a patient someday has a problem with amoxicillin during breast-feeding and I said it was ok to remove this alert.” –anonymous pharmacist
This statement is clinically accurate, risk-averse, and operationally paralyzing.
Clinically accurate: Yes, every alert is based on actual patient experience, including rare events
Risk-averse: if we have a tool to show risk, we should show ALL risk. Right?
Operationally paralyzing: Show ALL alerts? We would drown. Or we start ignoring them.
Here is a thought experiment.
On one side, place a pharmacist whose only responsibility is to avoid ANY medication side effects that have been described in the literature. In the extreme view, almost no medications would be prescribed, or at least the prescriber would have to read numerous descriptions of potential high, medium and lower risk interactions and alerts. This would certainly minimize medication risk. Or at least, as designers of the EHR, it would not be OUR FAULT since we showed ALL the alerts and the prescriber was warned!
On the other side, place a physician/prescriber ordering medications to treat patient conditions. This prescriber is burdened with seeing patients quickly and to take reasonable precautions. Consider that every medication alert, especially those with “medium to low risk” is a nuisance and a cognitive burden and contributes to burnout. In the extreme view, almost no alerts are helpful. Turn them all off.
How might be reconcile these opposite viewpoints in a productive conversation?
PRINCIPLE: Disconnect the fear of blame from leadership responsibility.
One way to address the pharmacist’s concerns of being individually blamed for decisions about alerts is to develop committee consensus. Let a group of prescribers, pharmacists and operational leaders consider and agree on reasonable settings for these alerts. If an alert is “overridden” 95% of the time by prescribers, then it is being ignored. As a governing body, we agree to eliminate all alerts with more than 90% override so that prescribers are more likely to pay attention to future alerts. It is much more difficult to attack a leadership consensus decision than an individual’s decision.
PRINCIPLE: What are the conflicting and balancing metrics in this decision?
Bring multiple viewpoints to committee discussion. One-dimensional decisions “we need to show ALL the alerts” versus “We need to show the alerts for EVERY med” are impractical when the decision affects multiple roles differently.
PRINCIPLE: Overall, how might we do the right thing for the patient?
Keeping the principle of “Do the right thing for the patient” can help clarify thinking. Lets say we present ALL the alerts. As a result, prescribers stop paying attention to all alerts and even critical alerts are now ignored. This is also not safe for the patient.
LESSON LEARNED
Using these principles, make some reasonable decisions. Stop being paralyzed with opposing points of view. Choose a reasonable path.
The Reasonable Path can feel scary! In the back of all of our heads, we worry about the lawsuit for the rare side effect or interaction that was actually listed on the package insert on page 27.
To find the reasonable path, use committee consensus to resolve a difficult decision.
Angry individuals with pitchforks and torches will have a much harder time attacking a thoughtful committee of diverse experts compared to attacking any single individual’s decision.
In this case, our EMO committee is formed of representatives from medicine, nursing, pharmacy, and information technology. The principles are clearly discussed, the viewpoints are respectfully debated. A leader helps develop a consensus or a well-informed decision. EMO has developed an energy and purpose and has achieve remarkable results, dismantling noisy, unhelpful alerts and improving the signal-to-noise ratio of others. Lets go disrupt the status quo. Lets go ruffle some feathers!
TRY THIS YOURSELF
What components of your committees are effective and less effective? What will you do about it?
Interesting post. Trying to strike the right balance between too many alerts and almost no alerts is difficult. Another challenge is making sure that the alerts are accurate. The breast feeding alert is a case in point. In IM, 98% of the time the patient is not actually breast feeding, but the alert fires because the breastfeeding box is erroneously checked, or the breastfeeding section has never been checked at all. Instead of having the provider click to ignore the alert, I would wonder about having the alert route the provider to the breastfeeding section of the EMR so the provider can click the correct box, and thereby stop future erroneous alerts.